Outdoor Point-Cloud Reconstruction Using Drone-Based Millimeter-Wave Systems

Abstract

Outdoor point-cloud reconstruction is challenging under poor illumination, adverse weather, and platform motion. Millimeter-wave (mmWave) sensing can operate in low-light and foggy environments, but produces sparse and noisy point clouds. We present an exploratory drone-based pipeline for small-scale outdoor reconstruction using frequency-modulated continuous-wave (FMCW) radar and constant false-alarm rate (CFAR) detection. A DJI M100 carries a 12-channel TI mmWave transceiver and DJI Guidance depth sensors that provide a reference PCD. The pipeline synchronizes both streams, detects range-Doppler objects, estimates azimuth and elevation, and converts detections into Cartesian points. In preliminary evaluation, the Chamfer-distance CDF reaches 0.9 near 1 m, while hovering increases the upper extent of back-to-back mmWave MSE by approximately fivefold, highlighting platform motion as a key challenge for airborne reconstruction.